基于部件退化动态耦合关系模型的数控系统多源信息融合可靠性研究

The reliability research of CNC (Computer Numerical Control) system is faced with the problems of fuzzy structure relationship of complex system and less valid sample data. This project aims to study reliability modeling and multi-source information fusion predictive evaluation methods of CNC system based on dynamic coupling model of component degradation. First, expand existing works, mining the reliability data of CNC system accumulated by the team, study weak links and the typical fault impacting factors, design new loading device, carry out reliability loading and enhancement tests, and explore the mechanism of performance degradation of the key components. Further study interaction relationship and evolution mechanism between performance degradation and physical of failure data, considering the distribution of degraded data and historical life and the similarity of component service, combined with hierarchical fault propagation directed graph, a new dynamic coupling model is established for CNC system component nodes and reliability attributes. Finally, a cooperative optimization algorithm for fusion weight is studied, and based on the new model and the reliability testing data of CNC system, a multi-source information fusion method for the reliability prediction and evaluation of CNC system is proposed. This research will be a useful supplement to the existing theory and technology of the reliability growth of complex mechatronics system.

数控系统可靠性研究面临复杂系统对象的部件结构关系模糊和有效样本数据少等难题。本项目旨在探索基于部件退化动态耦合关系模型的数控系统多源信息融合可靠性建模及预测评估方法。首先，拓展已有工作基础，挖掘课题组积累的数控系统可靠性实测数据，研究系统薄弱环节和典型故障影响因子，设计新型加载装置并开展加载和强化试验，探索关键部件性能退化机理；进而研究关键部件失效物理数据与性能退化之间的相互作用及演化关系，考量退化失效分布、历史寿命数据和部件服役相似度，结合分层故障传播有向图研究，构建数控系统部件节点与可靠性属性关联的动态耦合新模型；最后，基于新模型和数控系统可靠性实测数据等多源信息，研究融合权值的协同优化算法，提出一种多源信息融合的数控系统可靠性预测评估方法。研究成果将是对现有复杂机电系统产品可靠性增长理论和技术的有益补充。

数控系统是工业母机数控机床的大脑，然而我国高端机床和数控系统严重依赖进口、技术受制于人，产业链脆弱，究其主要原因在于可靠性与国际先进水平存在很大差距，因此探究数控系统部件退化关系并建立多源信息融合的可靠性模型，对提升我国数控系统可靠性水平，提高装备制造业的核心竞争力具有重要意义。.1) 提出了数控系统可靠性试验方法. 研究了系统薄弱环节和典型故障影响因子，设计了新型加载装置并开展加载和强化试验。首先采用故障模式、影响和危害性分析挖掘出数控系统的薄弱环节；之后基于典型加工工况，设计了驱动单元加速试验平台；编制了铣削加载谱；得到了适用于冲击载荷条件下的切削状态数据，为后续可靠性分析提供了数据支撑。.2) 构建了数控系统部件退化动态耦合模型. 建立了数控机床复杂网络模型与耦合关系模型，融合多种网络特征指标，基于层次分析法融合多种网络特征指标确定各个指标的权重，完成数控系统关键部件可靠性量化，研究了故障高风险传播路径和可能故障源的识别，分析数控系统故障传播机理，基于SIS-CA模型及目标边免疫策略，建立了考虑故障传播路径的预防性维修模型。.3) 提出了多源信息融合的数控系统可靠性预测评估方法. 开展了基于关联规则的数控系统故障数据挖掘，引入故障关联因子建立基于故障关联度的数控系统可靠性模型。提出了多源信息融合的数控系统可靠性模型，基于共轭分布和概率编码方法获得了故障时间数据和专家评判信息的先验分布，通过JSD方法评估了各先验分布的贡献度，对融合先验分布、极大似然估计和未融合先验分布的精确性进行了对比分析。